- The Inktelligence
- Posts
- The Inktelligence - March 3, 2025
The Inktelligence - March 3, 2025
From Shallow to Deep: LLMs as Your New Research Assistant
Hock Ng
March 03, 2025
In partnership with
Image generated with Midjourney
Can you believe that it’s already March? Well, as the weather warms up (slowly!) in the northern hemisphere, we’re also seeing the race between the AI companies heating up. In this issue, I’m going to take a look at the trend among the LLM providers giving us AI that goes through more “thinking” steps. Some call it “deep thinking” while others call it “deep research”.
First off, ChatGPT Plus users ($20/month) now have access to Deep Research. However, you’re only limited to 10 queries per month, so you’ll want to plan carefully. Pro users ($200/month) get 120 queries/month.
Just how good is Deep Research?
I’ll show you by way of an example.
Let’s say we’re interested in the AI Virtual Receptionist market and we want to study the competitors, market gaps, and how we can potentially enter this market. When you prompt Deep Research, it takes 2 steps. After you write the first prompt, Deep Research will come back to you with some clarification questions. The first prompt should describe in detail what you would like it to do.
Here’s an example of an outline that you can use for the first prompt:
Objective — the key question or insight you’re after (a competitive review, business plan, etc.)
Format — the sections that should be included in the output
Warnings — websites to avoid , limits on recency
Context — additional information about the task
Tip: If you need help, you can also use another ChatGPT conversation to come up with the information above.
In this case, my initial prompt specified four companies that surfaced as a result of another conversation with ChatGPT. So I asked for a detailed deep dive into these companies along with the pricing models and a SWOT analysis of each one. I asked for a McKinsey style report.
Here’s my prompt:
“I would like to perform market research on these companies, their addressable market, platform, technology, pricing models, and identify gaps in the market.
Dialzara: Provides AI virtual receptionists that handle incoming calls, capture crucial information, and transfer callers to the appropriate contact.
Smith.ai: Offers virtual receptionists capable of call handling, appointment scheduling, lead intake, and more, with 24/7 live agents to support plumbing companies.
Newo.ai: Specializes in AI receptionists for HVAC, electrical, plumbing, and appliance repair services, providing features like bookings, scheduling, cost estimations, and customer support.
Upfirst: Delivers 24/7 AI answering services tailored for plumbing companies, ensuring every call is answered, appointments are scheduled, and emergency calls are forwarded appropriately.
The output should be a comprehensive report like what McKinsey would deliver, including SWOT analysis for each company.”
Here are the clarification questions it then asked me: